Which is the best way to perform on the fly configurations?

1

Hello, we currently use GenieACS to configure about 200 CPEs in our network. Our goal is to migrate our provisioning system to GenieACS, and we need to configure about 16,000 CPEs.
Today, for the 200, we generate the configuration on the fly by reading a database to obtain the values ​​we need to configure and, using the “declare” function, we configure them on the CPEs.
I’ve read that other implementations do this by generating a configuration file on the fly and uploading it to the CPEs.
Which of these approaches do you think is the most scalable?
Do you know of another way to achieve this?
Thanks in advance.

2

See Example of a Provisioning Flow.

I’ve been provisioning using a variation of this code for 8-9 years.

High level overview of the process:

CPE Informs/Boots/Bootstraps to the ACS

Check if CPE has Provisioned tag. If so, exit the provision script

else

Call extension script to get provisioning information (username, password).

Apply config to cpe.

Reboot CPE.

3

Hi, our current approach is based on that example. We use tags to determine where we are in the provisioning process and retrieve the parameters to configure the CPE from our database using extension scripts.
However, so far we’ve only provisioned 200 CPEs this way and haven’t encountered massive configuration requests due to, for example, power outages at the CPEs. Therefore, we don’t know how scalable this approach is.
By the way, how are clients connected to a CPE monitored? Is it done via SNMP if possible or via TR-69? Is it done on demand when you want to query it or unattended for historical purposes? If you want to monitor other parameters, such as CPU usage, memory usage, or power received by the CPE, it’s recommended to do so via SNMP if possible or via TR-69. We currently monitor the CPE’s receiving power using SNMP on the OLT, but the CPU load on the OLT is too high, and we’ve seen cases where this is achieved using TR-69.
Thank you in advance.

4

You would not have any configuration requests from a CPE being power cycled. The customer would have to factory default the CPE for it to request its config again).

We don’t do any monitoring of our CPEs. Some CPEs have the ability to do SNMP. Some CPEs have the ability to report memory/cpu/etc via TR069/TR181.